Artificial Intelligence (AI), the design of technology with attributes that we traditionally associate with living beings, generally follows the broader scientific tradition of focusing on technical problems and their solutions within a relatively constrained framework. The cultural studies of science, on the other hand, insists that scientific work should be understood as it springs from and influences other cultures; phenomena, including the background of metaphors and assumptions that influence the way scientists do their work. In this thesis, I explore the possibilities for AI and the cultural studies of science to engage in a mutually beneficial alliance, by studying AI as a culturally situated activity and by using results of that study to generate novel technology.

Specifically, I focus on the design of autonomous agents, programs which are intended to represent a complete person, animal, or character. In the alternative AI tradition, these agents are created from a set of independent building blocks termed behaviors. A major open question is how these behaviors can by synthesized to create an agent with overall coherent behavior. I trace the problems in behavior integration to a strategy called atomization that AI shares with industrialization and psychiatric institutionalization. Atomization is the process of breaking agents into modular chunks with limited interaction and represents a catch-22 for AI; while this strategy is essential for building understandable code, it is fatal for creating agents that have the overall coherence we have come to associate with living beings.

I tackle this problem of integration by redefining the notion of agent. Instead of seeing agents as autonomous creatures with little reference to their sociocultural context, I suggest that agents can be thought of in the style of cultural studies as a form of communication between the agent's designer and the audience which will try to comprehend the agent's activity. With this metaphor as a basis, it becomes clear that we need to integrate, not the agent's internally defined code, but the way in which the agent presents itself to the user. Narrative psychology suggests that agents will be maximally comprehensible as intentional beings if they are structured to provide cues for narrative. I therefore build an agent architecture, the Expressivator, which provides support for narratively comprehensible agents, most notably by using behavioral transitions to link atomic behaviors into narrative sequences.

302 pages

Return to: SCS Technical Report Collection
School of Computer Science homepage

This page maintained by reports@cs.cmu.edu